Microsoft storage: a litany of failure
As Microsoft provokes a storm by canning Drive Extender, Jon Honeyball reflects on the company's long list of disasters in the storage arena
A deep, heartfelt sigh is all I can muster in response to the news that Microsoft’s Windows Home Server (WHS) team has decided to pull the Drive Extender technology from the next release of WHS, due later this year.
Apparently it’s too difficult, it has bugs, it doesn’t work under heavy loads, and it’s best that we don’t have it at all if it isn’t working properly (a view that has a certain logic to it).
The handling of this news has been somewhat catastrophic for the WHS team’s credibility. Apparently, we don’t actually need Drive Extender.
The handling of this news has been somewhat catastrophic for the WHS team’s credibility
In fact, it seems we’ve been asking, nay begging, Microsoft to pull it, so the firm is really just doing what we asked. No, really, that might seem a typical piece of Microsoft corporate double-speak to you and me, but it’s what we want.
Everyone laughed in amazement, then got angry and told Microsoft where it could shove all its future WHS products – and they said so in public, on the Microsoft forums.
Within hours came a second missive from the temple on top of the mountain. Actually, the work done to redesign Drive Extender from the existing version 1 to the newer, cleverer version 2, has failed. It’s buggy, it falls over, it loses your data, and Microsoft can’t fix it in time. You still don’t want or need it either, do you?
Dropping the ball
I’m not entirely surprised that Microsoft dropped the ball on this. Just look at its recent history with storage technologies.
WHS version 1 is a lovely thing, based on Server 2003 with all the unnecessary corporate gubbins pulled out. It was designed to run without a screen or keyboard.
It fitted nicely into home networks, and offered backup and recovery for desktop PCs. It offered a centralised store for multimedia content and streamed it on the network to any device that could consume it.
Part of that multimedia requirement was a storage solution that could just grow with the user’s collection over time, so Drive Extender was created, which basically faked a file store running on top of existing NTFS disk partitions.
And because it wasn’t tied to any underlying drive letters, you could add another hard disk and seamlessly grow your store without having to move everything around or reformat the disks. This was ideal for home users with a steadily growing storage requirement: when their disk gets full, just pop in an extra one.
Even better, Drive Extender added some extra tweaks that were genuinely useful. You could mark certain data as being important and WHS would ensure that, although Drive Extender made them appear that they were stored in one place, they would actually be stored across two physical drives, which protected you against a single disk failure.
This was no replacement for a full backup solution, but it at least kept the machine running and your data safe whenever a disk failed.
Such a solution was spot on for the home market, which was hardly surprising as that was where it was aimed in the first place. The implementation wasn’t without hiccups; there was a bug in the first release that allowed Drive Extender to lose data.
It took Microsoft months to fix it, which was the first indication that this wasn’t a technology it took very seriously.
Now scoot forward to the current betas and Microsoft has decided that Drive Extender is going to be useful to a wider audience. We might like it in the Small Business Server, too.
After all, we have similar storage needs there, which also grow over time. Buying an SBS server with a few empty drive bays gives us the room to grow in future. In order to make Drive Extender work well on a more business-orientated server, where the disks would be subjected to the heavy I/O loads of, for example, Exchange Server, it was decided to re-architect the product, and it’s this work that’s failed, because it now appears it does rather unpleasant things when under load.
So in typical Microsoft thinking, best pull the product from the whole platform and pretend that nothing happened, and that it’s all for our own good.